This specification relates to acoustic modeling.
Acoustic models may receive an acoustic sequence and generate a phone representation of the acoustic sequence. An acoustic modeling system can be used in, for example, a speech recognition system, e.g., in conjunction with a pronunciation model and a language model.